Objectives. We examined the role of community-level factors in explaining geographic variations in modern contraceptive use in 6 African countries.

Methods. We analyzed Demographic and Health Survey and contextual data sources with multilevel modeling techniques to identify factors contributing to geographic variations in women’s use of modern contraceptives.

Results. We found significant associations between several community-level factors and reported use of modern contraceptive methods. We also identified several pathways of influence between the community and the individual.

Conclusions. Aspects of a community’s sociocultural and economic environment appear to influence a woman’s use of modern contraceptive methods.

Many developing countries have substantial geographic variations in contraceptive use,1,2 although the factors shaping these variations are little understood. Previous studies suggested that variations in contraceptive use typically remain after accounting for individual and household factors.1,3 Contextual factors such as community-level cultural beliefs, the presence and quality of reproductive health services, the physical characteristics of the area, macroeconomic factors, and the presence of transport routes have been suggested as causes of geographic variations in contraceptive use.28 Studies of the determinants of contraceptive use, however, have focused on individual- and household-level factors.

The use of modern contraceptive methods (oral pills, condoms, intrauterine devices, sterilization, implants, and injectables) has traditionally been low in sub-Saharan Africa, but there is evidence of an increase during the past decade. During the 1990s, the self-reported use of modern contraceptives among married women increased in East Africa by 1% each year to 17% by 1999; in West Africa, it increased by 0.5% each year to 8% by 1999.9 Previous research concentrated on individual and household demographic and socioeconomic factors as determinants of the uptake of contraception. For example, associations between modern contraceptive use and parity,3,10,11 education,3,10,12 and household socioeconomic status11,13,14 are well established. However, evidence as to how community-level contextual factors influence variations in the use of contraceptive methods is limited.8,15 At the community level, studies of contraceptive use have focused on the influence of health service characteristics (for example, the influence of quality of care on contraceptive adoption).35,12

We examined geographic variations in modern contraceptive use in 6 countries from East (Kenya, Malawi, and Tanzania) and West (Burkina Faso, Ghana, and Ivory Coast) Africa, focusing on contextual factors that may influence contraceptive use. A greater understanding of the contextual factors associated with modern contraceptive use has the potential to inform the development of community-level programs aimed at increasing contraceptive use and to allow the targeting of programs to communities in need.

It is unlikely that characteristics of the health services in a community alone account for all geographic variation in contraceptive use. For example, contraceptive use may be indirectly influenced by economic development, through a relationship with access to health services7 or through its relationship with female autonomy and positive attitudes toward health service use.16 Some studies have examined the effects of other characteristics of the community, including the influence of levels of community economic development,2,6,7,17 levels of school participation,18,19 economic roles of children,20,21 and community fertility norms,4,6,22 on contraceptive use. We took a holistic view of contextual influences on modern contraceptive use, to provide insight into the role of the community in individual contraceptive behavior and contrast this across the 6 African countries.

Data

Six African countries were selected for analysis, from 2 regions: Ivory Coast, Burkina Faso, and Ghana in West Africa and Kenya, Malawi, and Tanzania in East Africa. The selection of neighboring countries allowed the identification of geographic variations in modern contraceptive use that may transcend political boundaries. We used individual and household data from the Demographic and Health Surveys (DHSs) conducted in the 6 study countries (Kenya, 1998; Malawi, 2000; Tanzania, 1999; Burkina Faso, 1998; Ivory Coast, 1998; and Ghana, 1998). Interviews were conducted with all women aged 15 to 49 years, regardless of marital status, in each sampled household, and data was collected on fertility and family planning in addition to demographic and socioeconomic data. The nonresponse rate for women ranged from 3% to 5% across the 6 countries. The sample of women in our analysis were of reproductive age who were not currently pregnant and were sexually active. The final sample sizes were: Tanzania, 3047 women; Ghana, 3784 women; Burkina Faso, 4960 women; Ivory Coast, 2437 women; Kenya, 6013 women; and Malawi, 10 291 women.

For each country, the DHS data contained identifiers for the communities (primary sampling units [PSUs]) and districts (administrative areas) where each participant resided. For our analysis, the PSUs are representative of the community. With the exception of the transportation infrastructure, which was measured at the district level, all contextual effects were considered at the community level. PSUs represent sampling blocks of approximately 20 to 30 households. The locations of each sampled PSU were obtained from ORC Macro (Calverton, Md) and were collected through global positioning system locators.23 International and subnational boundary data were obtained from the African Population Database (National Center for Geographic Information and Analysis, University of California, Santa Barbara). The data were plotted with the Arc-View Geographic Information System 8.2 (ESRI International, Redlands, Calif).

The majority of the contextual factors were taken from the DHS data, which entailed averaging individual data to the community level, thus producing derived community-level factors. Data were also obtained from Geographic Information System maps: rainfall (United Nations Environment Program, GEO Data Portal, ESRI International), habitat (ArcView World Map version 8.2, ESRI International), and road and rail networks (Digital Chart of the World, ArcView 8.2, ESRI International).

Analysis

The dependent variable for the analysis was a binary variable representing the use of a modern method of contraception (oral pill, intrauterine device, condom, female or male sterilization, implant, and injectable) at the time of the survey. Each of the DHS data sets had a hierarchical structure, with women living within PSUs, and located within districts. A multilevel modeling technique was employed to account for the hierarchical structure of the data and to facilitate the estimation of community- and district-level influences on modern contraceptive use. Multilevel modeling corrected the estimated standard errors to allow for clustering of individuals within areas.24 The multilevel regression models estimated variances in modern contraceptive use between the communities and between the districts. These variations represented the unexplained variation in contraceptive use that remained after accounting for the factors included in the models. A significant variance might have represented factors that influenced contraceptive use that were omitted from the models, either because they could not be quantified in a large survey or were absent from the data set, or a significant variance might reflect the poor measurement of some factors thought to influence contraceptive use. Separate multilevel logistic models were fitted for each of the 6 countries with the MLwiN version 2.0 software package (Bristol, England).25 Two levels of variance were considered, the PSU and the district.

The models are written as

(1)Yijk=πijk+εijkZijk,

where logeijk /[1− πijk]) = α + βX ijkT+ Ujk + VK, Yijk is a binary outcome (use of modern contraception) for individual i in PSU j in district k, and each Yijk are assumed to be independent Bernoulli random variables with the probability of use of modern contraception πijk = Pr(Yijk =1). Consequently, to correctly specify the binomial variation, Zijk denotes the square root of the expected binomial variance of πijk , and the variance of the individual residual term εijk is constrained to be 1. The outcome variable loge ijk /(1−πijk)) fitted in the model is the loge odds of use versus nonuse. This constrained the predicted values from the model to be between 0 and 1. The α is a constant, and the β is the vector of parameters corresponding to the vector of potential explanatory factors defined as Xijk. The PSU (level 2) residual term is defined as Ujk~N(0,σ 2u ), and the district (level 3) residual term is defined as Vk~N(0,σ2v ).

Our analysis was originally intended to measure the influence of several contextual factors on the use of modern contraceptive methods, including the health service environment, physical infrastructure, and prevailing cultural beliefs surrounding health care seeking. However, it proved difficult to obtain contextual data from many of the study settings because indicators were often measured differently across the 6 countries, limiting the standardized contextual-level data that was available across the 6 study countries. Data on the presence or quality of health services in each community were not available; thus we did not control for the potential influence of the health service environment in our analysis. Community-level data that were available for all 6 countries included community levels of fertility, desire for HIV testing, age at marriage, and transport infrastructure and habitat type of the district (Table 1). However, although these variables were available for all countries, they were not significant in the final models and thus are not presented in the analysis.

The same independent variables were entered into the models for all 6 countries (Table 1). At the first stage, individual and household factors were included in each model. All individual and household variables selected were significant in at least 1 of the 6 countries; thus none were excluded from the analysis. Next, standardized residuals, which compared the predicted level of contraceptive use for an area (from the model with no contextual variables) with the level of actual use (from the DHS data), were calculated for the districts in each country. These residuals were then mapped to visually identify any patterns in the districts that had much higher or lower contraceptive use than predicted. Finally, the contextual factors were added to the models. Only contextual-level factors that were significantly related to contraceptive use in at least 1 country were presented in the final model (Table 1). For each model, residual variation was estimated for the community (PSU) and district level. Changes in the district and community variances between the models were noted to test whether the addition of contextual factors to the analysis influenced the geographic variation in modern contraceptive use.

Tables 2 and 3 present the adjusted odds ratios for the modeling of modern contraceptive use. The results for the individual- and household-level covariates varied across the 6 countries. Women who did not want to have another child in the next 12 months were more likely to be using a modern method of contraception. Fecund women and women whose husbands approved of contraception were more likely to be using modern contraception in all 6 study countries. The fecundity measure uses the standard DHS variable (V623) for fecundity that differentiates between pregnant women, post-partum amenorrheic women, menopausal or infecund women, and fecund women. It is recoded to classify all nonfecund categories into 1 group. However, an additional category is added for those who currently state they are breastfeeding. With a the exception of Ghanaian women, women with a secondary education or higher were more likely to be using contraception than were women with no education. In the East African countries, women with no children were less likely to use modern contraception than were women with 3 or 4 children. The relationship between parity and modern contraceptive use was mixed in West Africa. In Burkina Faso, Malawi, Ivory Coast, and Tanzania, women who were never married had higher use of modern contraceptive methods than did married women or those in nonmarital unions. In Kenya, Malawi, and Tanzania, women aged 40 to 49 years were significantly less likely to be using contraception than women aged 30 to 39 years.

In all 6 countries, women who reported frequent discussion of family planning with their partners were more likely to be using contraception than were women who reported they never discussed family planning. In Kenya, Malawi, Burkina Faso, and Ivory Coast, women who reported being exposed to family planning information in the media were more likely to be using contraception. The relationship between modern contraceptive use and the household amenities index varied across the 6 countries, although higher odds of contraceptive use were generally seen in the wealthier households. The association between modern contraceptive use and religion was weak and only significant in 2 countries. In Malawi, Muslims were less likely to use modern contraception than were Catholics; in Ivory Coast, Protestants were significantly less likely to use modern contraception than were Catholics.

Figure 1 displays the level of contraceptive use in each of the districts (from the DHS data) and the district-level residuals for contraceptive use after accounting for the individual and household factors in the models. Modern contraceptive use varied greatly within each of the study countries. The maps for East Africa show that the districts with high or low residual variation in modern contraceptive use tended to cluster. Parts of southern Malawi, coastal Tanzania, and central Kenya typically had higher use of modern contraceptive methods than was predicted by the model. Conversely, parts of western Kenya had lower-than-predicted use of modern contraceptive methods. This clustering within countries suggests that there might be characteristics common to these districts that acted to shape modern contraceptive use and that these characteristics were geographically clustered. In West Africa, unexplained variation in contraceptive use was limited primarily to Burkina Faso. The districts with high or low modern contraceptive use did not transcend political boundaries, suggesting that the factors influencing the clustering were country specific.

Contextual factors were then included in each model. A range of factors were significant, but they varied across the 6 study countries. Level of approval of family planning among women in the community was the only factor significant in more than 1 country: Kenya, Malawi, Tanzania, and Ghana. Community socioeconomic factors were important in Tanzania (mean woman years of education for the community) and Burkina Faso (mean household amenities index in the community). In Malawi, religion was also significant (predominately Protestant communities compared with communities with a mixture of religions), and in Kenya, the percentage of male partners in the community who approved of family planning was significantly associated with the use of modern contraceptive methods. Finally, in Kenya, higher rainfall was associated with higher odds of modern contraceptive use.

Tables 2 and 3 show the estimates of district- and community-level variance that remained after all variables were included in the models. In 5 of the 6 countries, the district variation in modern contraceptive use was reduced by the addition of contextual factors; it remained significant only in Kenya and Malawi after the inclusion of all variables in the models. The addition of contextual variables to the models did not significantly reduce community-level variation in contraceptive use; after all variables were included in the models, significant community variation remained in Burkina Faso, Ghana, and Malawi. Therefore, the contextual variables included in the models were better predictors of variance in contraceptive use at the district level than at the smaller community level.

In the 6 countries studied, there were wide geographic variations in the use of modern contraceptive methods. Contextual factors, in addition to demographic and socioeconomic factors, have a significant role in creating geographic variations in modern contraceptive use. However, only the level of female approval of family planning in the community was found to be significantly associated with contraceptive use in both East and West Africa. In fact, in Ghana and Tanzania, the level of community approval of family planning had a larger effect on contraceptive use than did the perceived approval of the woman’s partner. This relationship suggests that a woman’s decision to adopt a modern method of contraceptive was strongly influenced by how she perceived other community members would judge her actions. Previous studies also showed that women may choose to adopt family planning, or indeed choose a particular method, as a result of the methods adopted by those in the community.26 The association between contraceptive use and the levels of approval by women of family planning in the community may also reflect various underlying community processes, such as prevailing cultural norms surrounding the expected roles of women.1,4,22

The contextual influences on modern contraceptive use identified varied by country, and only 2 significant contextual influences were identified in West Africa. However, together these 2 contextual influences reflect the importance of prevailing cultural norms, economic development, and very likely, service provision (although we were unable to account directly for this factor). The importance of community levels of socioeconomic development as an influence on contraceptive use is reflected in the significance of several variables. The significance of greater women’s education within the community in Tanzania highlights the effect of education on modern contraceptive use beyond the individual level. This is consistent with findings elsewhere that higher community education levels are associated with lower levels of fertility after accounting for individual education effects.27 Community education levels may be influencing prevailing norms regarding contraceptive decisionmaking through increasing levels of female autonomy or may reflect greater levels of economic development. The significance of community levels of the household amenities index in Burkina Faso also reflects a relationship between economic development and modern contraceptive use. The significance of religion in Malawi is another indicator of the influence of community-level cultural norms on modern contraceptive use. The significantly higher use of modern contraceptives in districts with higher rainfall levels in Kenya suggests that this factor is reflecting a more affluent socioeconomic climate, perhaps related to better agricultural production conditions.

The significant contextual factors were measured at the community level rather than the district level. However, they influenced district variation in modern contraceptive use more than they influenced community variation (they reduced the district-level random-effect term more than they reduced the community-level random-effect term). This indicates a greater similarity in contextual characteristics among communities within districts than among districts. This again suggests the importance of district-level development policies and cultural variation between districts and should act as motivation to include district-level characteristics in future analyses of contraceptive use.

Our findings of the ways in which aspects of the community influence a woman’s use of modern contraception can be used by family planning program managers to shape the development of family planning provision and promotion programs. For example, community health promotion can address cultural beliefs and customs to increase approval of family planning. In addition, the models can be used to identify areas where existing contraceptive use is above or below expectations. Areas with higher than expected use may be examples of good practice that providers and policymakers could learn from to improve policy and practice, and areas of lower use could be targeted for future interventions. The importance of community socioeconomic factors also lends support to development policies that indirectly influence community norms and behavior regarding family planning and the removal of economic barriers to service use.

A limitation of this research was the inability to control for the presence of health care services in each community; it is likely that the absence of such data was reflected in the remaining variation in modern contraceptive use. Also, some of the associations found with contextual-level variables may have acted as proxies for the effect of health services on contraceptive use. For example, more positive attitudes toward contraceptive use may be reflective of the presence of quality family planning services. The residual variation points to the need to improve the collection of health service data that can be linked to population-based data such as the DHS.

The results also should act as motivation to those involved in the routine collection of survey data to place a greater emphasis on the collection of community-level data. Given currently available data, the method we employed can be successfully applied to the examination of international comparisons of variations in health outcomes. Analysis of the variation within a single country would also be appropriate and simple to perform. The mechanisms through which the significant contextual factors influence individual contraceptive behavior need to be identified through further in-depth qualitative and quantitative research. Our results, however, have provided an important step toward understanding the numerous ways in which the health decisions made by individuals are influenced by the characteristics of the communities and districts in which they live.

Table
TABLE 1— Independent, Household, and Community Variables for Modeling of a Woman’s Use of a Modern Contraceptive Method: 6 African Countries, 1998–2000
TABLE 1— Independent, Household, and Community Variables for Modeling of a Woman’s Use of a Modern Contraceptive Method: 6 African Countries, 1998–2000
 Operational Definition
Individual and household variables
    Age of respondentSelf-reported age of respondent at time of survey: 15–19, 20–29, 30–39, 40–49
    ParityNumber of children ever born: 0, 1–2, 3–4, ≥ 5
    Self-reported fecundityaNot fecund, fecund, breastfeeding
    Self-reported fertility desireWants to have another child in next 12 months, does not want another child
    Marital status at time of surveyMarried or in nonmarital union, never married, formerly married, polygamous
    Educational attainmentHighest level attained: none, primary, secondary or higher
    ReligionCatholic, Protestant, Muslim, other
    Exposure to family planning informationRespondent has been exposed to family planning messages through newspapers, television, or radio: yes, no
    Partner’s residency statusPartner currently lives in the house, partner lives elsewhere
    Partner’s approval of family planningWoman’s report of partner’s attitude toward any family planning method: approves, does not approve
    Discussion of family planning with partnerAmount of discussion: never, once or twice, more than twice
    Household asset indexComposite index of household amenities, drinking water, toilet facility, floor material: categorized as 0, 1–2, 3–4, ≥ 5
    Urban residenceCurrent place of residence: urban, rural
Community variables
    Community level approval of family planning (husbands)Percentage of husbands in the community who are reported by their wives to approve of the use of any family planning method
    Community level approval of family planning (respondents)Percentage of respondents in the community who report that they approve of the use of any family planning method
    Community level of female educational attainmentThe mean number of years of female education per woman in the community
    Community household amenities indexMean amenities score among women in the community
    Religious compositionWhether > 50% of the respondents in the community are Muslim, Catholic, Protestant, other religions, or mixed
    Rainfall category of communityMillimeters of average annual rainfall (32 categories)
Other community variables tested but not in final models
    Mean number of children born per woman in the communityMean number of children ever born per woman in the community
    HIV testing in the communityPercentage of women in the community who report a desire to be tested for HIV
    Mean age at marriage for femalesMean age at marriage among respondents in the community
    Transport infrastructureKilometers of road per 1000 km2 in the district
    Habitat typeMain habitat type of the community: mangroves, tropical and subtropical moist broadleaf forests, tropical and subtropical grasslands, savannas, and shrublands

aThe fecundity measure uses the standard DHS variable (V623) for fecundity that differentiates between pregnant women, postpartum amenorrheic women, menopausal or infecund women, and fecund women. It is recorded to classify all nonfecund categories into 1 group. However, an additional category is added for those who currently state they are breastfeeding.

Table
TABLE 2— Adjusted Odds Ratios (ORs) for a Woman’s Use of a Modern Contraceptive Method: Kenya, Malawi, and Tanzania, 1998–2000
TABLE 2— Adjusted Odds Ratios (ORs) for a Woman’s Use of a Modern Contraceptive Method: Kenya, Malawi, and Tanzania, 1998–2000
 Kenya, OR (95% CI)Malawi, OR (95% CI)Tanzania, OR (95% CI)
Individual and household independent variables TQ1
Place of residence (rural)
    Urban1.02 (0.73, 1.42)1.14 (0.90, 1.45)1.56 (0.99, 2.45)
Want to have a child in the next 12 months (yes)
    Another answer9.40 (6.24, 14.16)6.90 (5.13, 9.28)4.84 (3.17, 7.37)
Fecundity (not fecund)a
    Breastfeeding0.71 (0.56, 0.91)1.00 (0.81, 1.23)0.48 (0.30, 0.77)
    Fecund2.84 (2.28, 3.53)2.04 (1.67, 2.49)1.74 (1.14, 2.65)
Age group (30–39)
    15–190.84 (0.58, 1.22)1.29 (1.01, 1.66)1.47 (0.85, 2.55)
    20–291.18 (0.96, 1.44)1.53 (1.31, 1.79)1.88 (1.35, 2.62)
    40–490.72 (0.59, 0.90)0.62 (0.53, 0.74)0.61 (0.41, 0.90)
Parity (3–4)
    None0.17 (0.12, 0.25)0.21 (0.15, 0.29)0.27 (0.15, 0.48)
    1–20.79 (0.65, 0.97)0.57 (0.49, 0.66)0.78 (0.56, 1.08)
    ≥ 51.20 (0.98, 1.47)1.56 (1.33, 1.82)1.09 (0.77, 1.56)
Marital status (married/nonmarital union)
    Never married1.43 (0.99, 2.07)2.07 (1.45, 2.97)3.38 (1.76, 6.51)
    Formerly married1.34 (0.95, 1.91)2.18 (1.68, 2.84)2.61 (1.48, 4.61)
    Polygamous0.84 (0.66, 1.07)0.96 (0.82, 1.14). . .
Respondent’s education (none)
    Primary1.30 (1.01, 1.67)1.07 (0.93, 1.22)1.30 (0.91, 1.84)
    Secondary/higher1.49 (1.13, 1.98)1.71 (1.39, 2.11)1.75 (1.49, 2.01)
Religion (Catholic)
    Protestant0.89 (0.75, 1.05)0.93 (0.82, 1.06)1.08 (0.74, 1.56)
    Muslim0.72 (0.47, 1.11)0.80 (0.63, 0.99)0.92 (0.65, 1.32)
    No religion/other0.73 (0.43, 1.25)0.63 (0.37, 1.05)0.95 (0.45, 2.00)
Partner living in household (yes)
    Staying elsewhere0.80 (0.67, 0.97)0.77 (0.64, 0.92)0.55 (0.34, 0.88)
Partner approves of family planning (disapproves)
    Approves3.49 (2.73, 4.46)3.59 (2.93, 4.39)2.28 (1.54, 3.39)
    Does not know1.04 (0.68, 1.57)0.46 (0.28, 0.76)0.53 (0.25, 1.15)
Discuss family planning with partner (never)
    Once or twice1.54 (1.21, 1.98)1.54 (1.28, 1.84)2.14 (1.28, 3.59)
    More than twice2.07 (1.60, 2.68)2.44 (2.04, 2.93)3.55 (2.15, 5.85)
Exposure to family planning in media (no)
    Yes1.32 (1.13, 1.54)1.32 (1.16, 1.50)1.26 (0.96, 1.64)
Household amenities index (≥ 5)b
    00.33 (0.21, 0.51)0.55 (0.41, 0.74)0.50 (0.17, 1.48)
    1–20.63 (0.49, 0.81)0.63 (0.51, 0.79)0.62 (0.37, 1.04)
    3–40.71 (0.57, 0.89)0.73 (0.60, 0.89)0.99 (0.71, 1.39)
Contextual independent variables
Mean female years of education in community1.08 (0.98, 1.18)0.96 (0.90, 1.03)1.28 (1.07, 1.52)
Mean household amenities index (community)1.06 (0.87, 1.30)1.00 (0.84, 1.20)0.91 (0.59, 1.40)
Dominant religion in communityc (mixed)
    Catholic0.74 (0.52, 1.05)1.18 (0.87, 1.61)0.76 (0.49, 1.56)
    Protestant0.86 (0.64, 1.15)1.39 (1.11, 1.74)0.91 (0.53, 1.56)
    Muslim1.03 (0.61, 1.76)1.21 (0.89, 1.66)0.98 (0.63, 1.53)
    None/other1.02 (0.34, 3.12). . .1.32 (0.46, 3.77)
Female approval of family planning (community), %2.67 (1.18, 6.00)2.84 (1.13, 7.19)12.01 (2.67, 54.0)
Partner approval of family planning (community), %2.07 (1.27, 3.36)1.46 (0.85, 2.51)0.81 (0.32, 2.10)
Rainfall category of community1.04 (1.01, 1.06)1.02 (0.97, 1.06)0.98 (0.96, 1.0)
Variance
District, B (SE)0.149 (0.052)0.034 (0.017)0.131 (0.069)
Community, B (SE)0.074 (0.042)0.128 (0.030)0.089 (0.075)

Note. CI = confidence interval; SE = standard error. Parenthetical items are reference category.

aThe fecundity measure uses the standard DHS variable (V623) for fecundity that differentiates between pregnant women, postpartum amenorrheic women, menopausal or infecund women, and fecund women. It is recorded to classify all nonfecund categories into 1 group. However, an additional category is added for those who currently state they are breastfeeding.

bComposite index of household amenities, drinking water, toilet facility, floor material: categorized as 0, 1–2, 3–4, ≥ 5.

cDefined as religion in community to which more than 50% of the females in the community belonged.

Table
TABLE 3— Adjusted Odds Ratios (ORs) for a Woman’s Use of a Modern Contraceptive Method: Ivory Cost, Burkina Faso, and Ghana, 1998
TABLE 3— Adjusted Odds Ratios (ORs) for a Woman’s Use of a Modern Contraceptive Method: Ivory Cost, Burkina Faso, and Ghana, 1998
 Burkina Faso, OR (95% CI)Ghana, OR (95% CI)Ivory Coast, OR (95% CI)
Individual and household independent variables
Place of residencea
    Urban. . .1.22 (0.84, 1.77). . .
    Abidjan, intervention zone. . .. . .1.83 (0.73, 4.58)
    Abidjan, nonintervention zone. . .. . .2.44 (0.90, 6.57)
    Other cities, intervention zone0.75 (0.23, 2.43). . .1.62 (0.87, 3.02)
    Other cities, nonintervention zone1.10 (0.25, 4.81). . .0.95 (0.51, 1.77)
    Rural intervention zone1.73 (0.35, 8.59). . .. . .
    Rural nonintervention zone1.26 (0.26, 6.01). . .. . .
Want to have a child in the next 12 months (yes)
    Another answer5.76 (3.21, 10.33)4.98 (2.98, 8.30)4.12 (2.49, 6.82)
Fecundity (not fecund)b
    Breastfeeding0.81 (0.43, 1.54)1.37 (0.78, 2.40)0.79 (0.32, 1.95)
    Fecund2.60 (1.43, 4.72)3.89 (2.33, 6.50)3.58 (1.60, 8.00)
Age group (30–39)
    15–191.07 (0.60, 1.91)1.24 (0.71, 2.14)1.26 (0.72, 2.21)
    20–291.07 (0.71, 1.61)1.08 (0.77, 1.50)1.81 (1.17, 2.80)
    40–490.68 (0.43, 1.08)0.75 (0.53, 1.05)0.86 (0.50, 1.46)
Parity (3–4)
    None1.31 (0.70, 2.45)0.82 (0.49, 1.40)0.55 (0.31, 0.97)
    1–21.17 (0.78, 1.75)0.71 (0.51, 0.99)0.55 (0.35, 0.88)
    ≥ 51.26 (0.82, 1.94)1.15 (0.82, 1.61)0.89 (0.54, 1.47)
Marital status (married/nonmarital union)
    Never married5.24 (2.56, 10.72)1.58 (0.79, 3.18)3.44 (1.59, 7.42)
    Formerly married1.67 (0.76, 3.64)1.35 (0.72, 2.51)2.10 (0.95, 4.64)
    Polygamous0.84 (0.59, 1.19)0.89 (0.64, 1.25)1.68 (1.06, 2.66)
Respondent’s education (none)
    Primary1.52 (1.05, 2.21)0.96 (0.67, 1.38)1.53 (1.07, 2.19)
    Secondary/higher2.52 (1.71, 3.72)1.20 (0.87, 1.66)2.20 (1.48, 3.27)
Religion (Catholic)
    Protestant1.23 (0.74, 2.05)0.89 (0.63, 1.25)0.59 (0.41, 0.86)
    Muslim0.81 (0.60, 1.11)1.07 (0.63, 1.80)0.96 (0.65, 1.42)
    No religion/other0.82 (0.43, 1.56)0.95 (0.59, 1.51)0.68 (0.44, 1.05)
Partner living in household (yes)
    Staying elsewhere0.75 (0.41, 1.35)0.67 (0.50, 0.90)1.29 (0.75, 2.21)
Partner approves of family planning (disapproves)
    Approves2.39 (1.50, 3.82)2.58 (1.60, 4.18)2.39 (1.35, 4.23)
    Does not know0.38 (0.18, 0.82)0.90 (0.50, 1.62)0.45 (0.17, 1.14)
Discuss family planning with partner (never)
    Once or twice2.99 (1.95, 4.58)2.29 (1.53, 3.44)1.32 (0.66, 2.62)
    > Twice3.80 (2.54, 5.68)4.87 (3.26, 7.27)2.16 (1.08, 4.30)
Exposure to family planning in media (no)
    Yes1.44 (1.06, 1.96)1.09 (0.83, 1.42)1.44 (1.06, 1.96)
Household amenities index (≥ 5)c
    00.86 (0.20, 3.63)0.11 (0.01, 0.91)0.06 (0.01, 0.52)
    1–21.12 (0.57, 2.20)0.72 (0.40, 1.28)0.06 (0.01, 0.53)
    3–40.97 (0.54, 1.74)0.82 (0.56, 1.20)0.23 (0.09, 0.59)
Contextual independent variables
Mean female years of education in community0.93 (0.78, 1.11)1.03 (0.87, 1.22)0.95 (0.74, 1.23)
Mean household amenities index (community)2.24 (1.08, 4.64)0.90 (0.59, 1.36)0.65 (0.35, 1.20)
Dominant religion in communityd (mixed)
    Catholic0.95 (0.57, 1.56)1.10 (0.60, 2.04)0.97 (0.63, 1.51)
    Protestant1.34 (0.12, 14.42)1.17 (0.77, 1.77)2.25 (0.44, 11.56)
    Muslim0.98 (0.65, 1.48)1.03 (0.54, 1.97)0.79 (0.51, 1.23)
    None/traditional/other0.82 (0.35, 1.94)1.19 (0.60, 2.35)0.90 (0.41, 2.00)
Female approval of family planning (community), %0.97 (0.23, 4.01)3.80 (1.57, 9.21)2.51 (0.65, 9.73)
Partner approval of family planning (community), %2.21 (0.74, 6.60)0.77 (0.40, 1.50)0.54 (0.22, 1.33)
Rainfall category of community1.01 (0.92, 1.12)1.07 (0.97, 1.18)0.94 (0.86, 1.03)
Variance
District, B (SE)0.338 (0.187)0.014 (0.022)0.199 (0.123)
Community, B (SE)0.188 (0.091)0.186 (0.093)0.062 (0.075)

Note. OR = odds ratio; CI = confidence interval; SE = standard error.

aFor place of residence, the reference category for Burkina Faso is Ouagadougou; for Ghana and Ivory Coast, the reference category is rural. For Ivory Coast, urban areas are divided into 4 groups: zones in Abidjan and other cities that received a health and family planning intervention and those did not. For Burkina Faso, the comparison groups are divided into 4: cities other than Ouagadougou that received a health and family planning intervention, other cities that did not, rural areas that received the intervention, and those that did not.

b The fecundity measure uses the standard DHS variable (V623) for fecundity that differentiates between pregnant women, postpartum amenorrheic women, menopausal or infecund women, and fecund women. It is recorded to classify all nonfecund categories into 1 group. However, an additional category is added for those who currently state they are breastfeeding.

cComposite index of household amenities, drinking water, toilet facility, floor material: categorized as 0, 1–2, 3–4, ≥ 5.

dDefined as religion in community to which more than 50% of the females in the community belonged.

This study was funded by the Economic and Social Research Council, UK (grant R000239664).

Human Participant Protection No protocol approval was needed for this study because it analyzed secondary data.

References

1. Amin S, Basu AM, Stephenson R. Spatial variation in contraceptive use in Bangladesh: looking beyond the borders. Demography.2002;39:251–268. Crossref, MedlineGoogle Scholar
2. National Research Council. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: National Academy Press; 1993. Google Scholar
3. Tuoane ‘M, Diamond I, Madise N. Use of family planning services in Lesotho: the importance of quality of care and access. Afr Popul Stud.2003;18:105–132. Google Scholar
4. Bongaarts J, Bruce J. The causes of unmet need for contraception and the social content of services. Stud Fam Plann.1995;26:57–75. Crossref, MedlineGoogle Scholar
5. Tsui AO, Ochoa LH. Service proximity as a determinant of contraceptive behaviour: evidence from cross-national studies of survey data. In: Phillips JF, Ross JA, eds. Family Planning Programmes and Fertility. Oxford, England: Clarendon Press; 1992:222–256. Google Scholar
6. Nazzar A, Adongo PB, Binka FN, Phillips JE, Debpuur C. Developing a culturally appropriate family planning program for the Navrongo experiment. Stud Fam Plann.1995;26:307–324. Crossref, MedlineGoogle Scholar
7. Diez-Roux AV. Bringing context back into epidemiology: variables and fallacies in multilevel analysis. Am J Public Health.1998;88:216–222. LinkGoogle Scholar
8. Stephenson RB, Tsui AO. Contextual influences on reproductive health service use in Uttar Pradesh, India. Stud Fam Plann.2002;3:309–320. CrossrefGoogle Scholar
9. United Nations Population Division. World Contraceptive Use 2003. New York, NY: United Nations; 2003. Google Scholar
10. Curtis SL, Neitzel K. Contraceptive Knowledge, Use, and Sources. Calverton, Md: Macro International; 1996. Demographic and Health Surveys Comparative Studies, No. 19. Google Scholar
11. Clements S, Madise N. Who is being served least by family planning providers? A study of modern contraceptive use in Ghana, Tanzania and Zimbabwe. Afr J Reprod Health.2004;8:124–136. Crossref, MedlineGoogle Scholar
12. Oliver R. Contraceptive Use in Ghana. The Role of Service Availability, Quality and Price. Washington, DC: World Bank; 1995. Living Standards Measurement Study Working Paper No. 111. Google Scholar
13. Thomas D, Maluccio J. Contraceptive Choice, Fertility, and Public Policy in Zimbabwe. Washington DC: World Bank; 1995. Living Standards Measurement Study Working Paper No. 109. Google Scholar
14. Guilkey DK, Jayne S. Fertility transition in Zimbabwe: determinants of contraceptive use and method choice. Popul Stud.1997;51:173–189. CrossrefGoogle Scholar
15. Stephenson R, Tsui AO. Contextual influences on reproductive wellness in northern India. Am J Public Health. 93:1820–1829. LinkGoogle Scholar
16. Alan Guttmacher Institute. Hopes and Realities: Closing the Gap Between Women’s Aspirations and Their Reproductive Experiences. New York, NY: Alan Guttamacher Institute; 1998. Google Scholar
17. Saha T. The Effect of Settlement Zones and Economic Diversity on Reproductive Behaviour in West Africa. West Africa Spatial Analysis Prototype Exploratory Analysis. Calverton, Md: Macro International; 1998. Google Scholar
18. DeGraff D, Bilsborrow RE, Guilkey DK. Community-level determinants of contraceptive use in the Philippines: a structural analysis. Demography.1997;34:385–398. Crossref, MedlineGoogle Scholar
19. Chacko E. Women’s use of contraception in rural India: a village-level study. Health Place.2001;7:197–208. Crossref, MedlineGoogle Scholar
20. Entwisle B, Casterline JB, Sayed AA. Villages as contexts for contraceptive behavior in rural Egypt. Am Sociol Rev.1989;54:1019–1034. CrossrefGoogle Scholar
21. Entwisle B, Mason WM. Multilevel effects of socioeconomic development and family planning programs on children ever born. Am J Sociol.1985;91:616–649. CrossrefGoogle Scholar
22. Nsemuklia GB, Phiri DS, Diallo HM, Banda SS, Benaya WK, Kitahara N. A Study of Factors Associated With Maternal Mortality in 1998. Lusaka, Zambia: Ministry of Health, United Nations Population Fund, Central Board of Health, and University of Zambia; 1999. Google Scholar
23. Montana L, Spencer J. Incorporating Geographic Information Into MEASURE Surveys: A Field Guide to GPS Data Collection. Calverton, Md: Macro International; 2001. Google Scholar
24. Goldstein H. Multilevel Statistical Methods. London, England: Arnold; 2003. Google Scholar
25. Institute of Education, University of London. Multilevel modeling project. Available at: http://multilevel.ioe.ac.uk/index.html. Accessed November 11, 2006. Google Scholar
26. Potter JE. The persistence of outmoded contraceptive regimes: the case of Mexico. Popul Dev Rev.1999; 25:703–739. CrossrefGoogle Scholar
27. Kravdal O. Education and fertility in sub-Saharan Africa: individual and community effects. Demography.2002;39:233–250. MedlineGoogle Scholar

Related

No related items

TOOLS

SHARE

ARTICLE CITATION

Rob Stephenson, PhD, Angela Baschieri, PhD, Steve Clements, PhD, Monique Hennink, PhD, and Nyovani Madise, PhDRob Stephenson and Monique Hennink are with the Hubert Department of Global Health, Rollins School of Public Health, Emory University, Atlanta, Ga. Angela Baschieri is with the Center for Population Studies, London School of Hygiene and Tropical Medicine, London, England. Steve Clements is with the Center for Sexual Health Research, Department of Psychology, University of Southampton, Southampton, England. Nyovani Madise is with the African Population and Health Research Center, Nairobi, Kenya. “Contextual Influences on Modern Contraceptive Use in Sub-Saharan Africa”, American Journal of Public Health 97, no. 7 (July 1, 2007): pp. 1233-1240.

https://doi.org/10.2105/AJPH.2005.071522

PMID: 17538071